@inerrata-corporation/errata 2.0.2-dev.503 → 2.0.2-dev.509

package/consolidate-worker.mjs

@@ -16562,6 +16562,19 @@ var SqliteGraphStore = class {
     const rows = this.db.prepare("SELECT from_id, to_id, type FROM edges WHERE valid_to IS NULL").all();
     return rows.map((r) => ({ from: r.from_id, to: r.to_id, type: r.type }));
   }
+  liveNodeIds(ids) {
+    const live = /* @__PURE__ */ new Set();
+    const CHUNK = 900;
+    for (let i = 0; i < ids.length; i += CHUNK) {
+      const slice = ids.slice(i, i + CHUNK);
+      if (slice.length === 0) continue;
+      const rows = this.db.prepare(
+        `SELECT id FROM nodes WHERE valid_to IS NULL AND id IN (${slice.map(() => "?").join(",")})`
+      ).all(...slice);
+      for (const r of rows) live.add(r.id);
+    }
+    return live;
+  }
   /**
    * Live edges WITH their endpoint labels and ids, resolved in ONE join.
    *

package/errata.mjs

@@ -17237,6 +17237,19 @@ CREATE INDEX IF NOT EXISTS nodes_relpath ON nodes(json_extract(attrs_json, '$.re
         const rows = this.db.prepare("SELECT from_id, to_id, type FROM edges WHERE valid_to IS NULL").all();
         return rows.map((r) => ({ from: r.from_id, to: r.to_id, type: r.type }));
       }
+      liveNodeIds(ids) {
+        const live = /* @__PURE__ */ new Set();
+        const CHUNK = 900;
+        for (let i2 = 0; i2 < ids.length; i2 += CHUNK) {
+          const slice = ids.slice(i2, i2 + CHUNK);
+          if (slice.length === 0) continue;
+          const rows = this.db.prepare(
+            `SELECT id FROM nodes WHERE valid_to IS NULL AND id IN (${slice.map(() => "?").join(",")})`
+          ).all(...slice);
+          for (const r of rows) live.add(r.id);
+        }
+        return live;
+      }
       /**
        * Live edges WITH their endpoint labels and ids, resolved in ONE join.
        *
@@ -19462,10 +19475,12 @@ function parseIntentFences(text) {
 }
 function searchByIntent(store, intent, opts) {
   const limit = opts?.limit ?? 5;
-  const minCoverage = opts?.minCoverage ?? 0.25;
+  const minCoverage = opts?.minCoverage ?? DEFAULT_MIN_COVERAGE;
   const want = new Set(retrievalTokens(intent));
   if (want.size < MIN_INTENT_TOKENS) return [];
-  const hits = [];
+  const candidates = [];
+  const df = /* @__PURE__ */ new Map();
+  let corpus = 0;
   for (const label of RECALLABLE) {
     for (const node2 of store.findNodesByLabel(label)) {
       if (node2.attrs["mergedInto"] !== void 0) continue;
@@ -19474,19 +19489,30 @@ function searchByIntent(store, intent, opts) {
       }
       const bag = retrievalTokens(node2.description ?? "");
       if (bag.length === 0) continue;
+      corpus++;
       const matched = bag.filter((t) => want.has(t));
-      if (matched.length === 0) continue;
-      const coverage = matched.length / want.size;
-      if (coverage < minCoverage) continue;
-      hits.push({ node: node2, coverage, matched });
+      for (const t of matched) df.set(t, (df.get(t) ?? 0) + 1);
+      if (matched.length > 0) candidates.push({ node: node2, matched });
     }
   }
+  const idf = (t) => Math.log(1 + corpus / (1 + (df.get(t) ?? 0)));
+  let denom = 0;
+  for (const t of want) if ((df.get(t) ?? 0) > 0) denom += idf(t);
+  if (denom === 0) return [];
+  const hits = [];
+  for (const { node: node2, matched } of candidates) {
+    let num2 = 0;
+    for (const t of matched) num2 += idf(t);
+    const coverage = num2 / denom;
+    if (coverage < minCoverage) continue;
+    hits.push({ node: node2, coverage, matched });
+  }
   hits.sort(
     (a, b) => b.coverage - a.coverage || (a.node.description?.length ?? 0) - (b.node.description?.length ?? 0)
   );
   return hits.slice(0, limit);
 }
-var RECALLABLE, MIN_INTENT_TOKENS;
+var RECALLABLE, MIN_INTENT_TOKENS, DEFAULT_MIN_COVERAGE;
 var init_intent = __esm({
   "../../packages/local-graph/src/intent.ts"() {
     "use strict";
@@ -19494,6 +19520,7 @@ var init_intent = __esm({
     init_design_problem();
     RECALLABLE = ["Problem", "Solution", "RootCause", "Pattern"];
     MIN_INTENT_TOKENS = 2;
+    DEFAULT_MIN_COVERAGE = 0.2;
   }
 });
 
@@ -21762,6 +21789,7 @@ __export(src_exports2, {
   CITABLE_PRIOR_LABELS: () => CITABLE_PRIOR_LABELS,
   CODE_REACH_EDGES: () => CODE_REACH_EDGES,
   CREDIT_FAMILY: () => CREDIT_FAMILY,
+  DEFAULT_MIN_COVERAGE: () => DEFAULT_MIN_COVERAGE,
   FIX_CANDIDATE_ASK_LIMIT: () => FIX_CANDIDATE_ASK_LIMIT,
   JOINT_COMMUNITY_EDGES: () => JOINT_COMMUNITY_EDGES,
   JOINT_COMMUNITY_LABELS: () => JOINT_COMMUNITY_LABELS,
@@ -37781,9 +37809,13 @@ function* walkSource(dir) {
 function listSourceFiles(root) {
   return gitSourceFiles(root) ?? walkSource(root);
 }
-function findStaleFiles(store, rootPath, workspaceId2) {
+async function findStaleFiles(store, rootPath, workspaceId2) {
   const stale = [];
+  const pending = [];
+  const allTargets = [];
+  let scanned = 0;
   for (const abs of listSourceFiles(rootPath)) {
+    if (++scanned % SCAN_YIELD_EVERY === 0) await new Promise((r) => setImmediate(r));
     let mtimeMs;
     try {
       mtimeMs = statSync2(abs).mtimeMs;
@@ -37800,11 +37832,21 @@ function findStaleFiles(store, rootPath, workspaceId2) {
       const edges = store.outEdges(fnode.id, ["DEFINES", "CONTAINS"]);
       if (edges.length === 0) {
         stale.push(abs);
-      } else if (edges.some((e) => !isLive(store.getNode(e.to)))) {
-        stale.push(abs);
+      } else {
+        const targets = edges.map((e) => e.to);
+        pending.push({ abs, targets });
+        allTargets.push(...targets);
       }
     }
   }
+  const live = /* @__PURE__ */ new Set();
+  for (let i2 = 0; i2 < allTargets.length; i2 += LIVENESS_CHUNK) {
+    for (const id of store.liveNodeIds(allTargets.slice(i2, i2 + LIVENESS_CHUNK))) live.add(id);
+    if (i2 + LIVENESS_CHUNK < allTargets.length) await new Promise((r) => setImmediate(r));
+  }
+  for (const { abs, targets } of pending) {
+    if (targets.some((id) => !live.has(id))) stale.push(abs);
+  }
   return stale;
 }
 function isLive(n) {
@@ -37820,7 +37862,7 @@ function findStrandedSymbols(store) {
   return stranded;
 }
 async function reconcileStaleFiles(store, rootPath, workspaceId2) {
-  const stale = findStaleFiles(store, rootPath, workspaceId2);
+  const stale = await findStaleFiles(store, rootPath, workspaceId2);
   if (stale.length === 0) return 0;
   let total = 0;
   const BATCH = 20;
@@ -37831,13 +37873,15 @@ async function reconcileStaleFiles(store, rootPath, workspaceId2) {
   }
   return total;
 }
-var IGNORED, SOURCE_RE;
+var IGNORED, SOURCE_RE, SCAN_YIELD_EVERY, LIVENESS_CHUNK;
 var init_reconcile = __esm({
   "src/reconcile.ts"() {
     "use strict";
     init_src11();
     IGNORED = /[\\/](?:\.git|node_modules|\.errata|dist|__pycache__)(?:[\\/]|$)/;
     SOURCE_RE = /\.(ts|tsx|js|jsx|mjs|cjs|py)$/i;
+    SCAN_YIELD_EVERY = 100;
+    LIVENESS_CHUNK = 4e3;
   }
 });
 
@@ -53797,7 +53841,7 @@ function startLoopLagMonitor(thresholdMs = 1e3) {
 }
 
 // src/engine.ts
-var DAEMON_VERSION = true ? "2.0.2-dev.503" : "2.0.0-alpha.0";
+var DAEMON_VERSION = true ? "2.0.2-dev.509" : "2.0.0-alpha.0";
 var IGNORED_PATH = /[\\/](?:\.git|node_modules|\.errata|\.claude|\.codex|\.cursor|\.turbo|\.next|dist|coverage|test-results|playwright-report|__pycache__)(?:[\\/]|$)/;
 var IGNORED_NOISE = /castalia\.db|eventlog\.sqlite|turn-cursors/;
 var GIT_OP_MUTE_MS = 4e3;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inerrata-corporation/errata",
-  "version": "2.0.2-dev.503",
+  "version": "2.0.2-dev.509",
   "description": "errata - local-first observation engine for AI coding agents",
   "type": "module",
   "bin": {

package/pass-worker.mjs

@@ -24889,6 +24889,19 @@ var SqliteGraphStore = class {
     const rows = this.db.prepare("SELECT from_id, to_id, type FROM edges WHERE valid_to IS NULL").all();
     return rows.map((r) => ({ from: r.from_id, to: r.to_id, type: r.type }));
   }
+  liveNodeIds(ids) {
+    const live = /* @__PURE__ */ new Set();
+    const CHUNK = 900;
+    for (let i2 = 0; i2 < ids.length; i2 += CHUNK) {
+      const slice = ids.slice(i2, i2 + CHUNK);
+      if (slice.length === 0) continue;
+      const rows = this.db.prepare(
+        `SELECT id FROM nodes WHERE valid_to IS NULL AND id IN (${slice.map(() => "?").join(",")})`
+      ).all(...slice);
+      for (const r of rows) live.add(r.id);
+    }
+    return live;
+  }
   /**
    * Live edges WITH their endpoint labels and ids, resolved in ONE join.
    *
@@ -26708,9 +26721,15 @@ function* walkSource(dir) {
 function listSourceFiles(root) {
   return gitSourceFiles(root) ?? walkSource(root);
 }
-function findStaleFiles(store2, rootPath, workspaceId) {
+var SCAN_YIELD_EVERY = 100;
+var LIVENESS_CHUNK = 4e3;
+async function findStaleFiles(store2, rootPath, workspaceId) {
   const stale = [];
+  const pending = [];
+  const allTargets = [];
+  let scanned = 0;
   for (const abs of listSourceFiles(rootPath)) {
+    if (++scanned % SCAN_YIELD_EVERY === 0) await new Promise((r) => setImmediate(r));
     let mtimeMs;
     try {
       mtimeMs = statSync2(abs).mtimeMs;
@@ -26727,18 +26746,25 @@ function findStaleFiles(store2, rootPath, workspaceId) {
       const edges = store2.outEdges(fnode.id, ["DEFINES", "CONTAINS"]);
       if (edges.length === 0) {
         stale.push(abs);
-      } else if (edges.some((e) => !isLive(store2.getNode(e.to)))) {
-        stale.push(abs);
+      } else {
+        const targets = edges.map((e) => e.to);
+        pending.push({ abs, targets });
+        allTargets.push(...targets);
       }
     }
   }
+  const live = /* @__PURE__ */ new Set();
+  for (let i2 = 0; i2 < allTargets.length; i2 += LIVENESS_CHUNK) {
+    for (const id of store2.liveNodeIds(allTargets.slice(i2, i2 + LIVENESS_CHUNK))) live.add(id);
+    if (i2 + LIVENESS_CHUNK < allTargets.length) await new Promise((r) => setImmediate(r));
+  }
+  for (const { abs, targets } of pending) {
+    if (targets.some((id) => !live.has(id))) stale.push(abs);
+  }
   return stale;
 }
-function isLive(n) {
-  return n != null && (n.validTo ?? null) === null;
-}
 async function reconcileStaleFiles(store2, rootPath, workspaceId) {
-  const stale = findStaleFiles(store2, rootPath, workspaceId);
+  const stale = await findStaleFiles(store2, rootPath, workspaceId);
   if (stale.length === 0) return 0;
   let total = 0;
   const BATCH = 20;